Positioning path memory and breakpoint operation method and system for a robot group

Abstract

The application discloses a positioning path memory and breakpoint operation method and system of a robot group, and relates to the technical field of robot positioning and navigation. The method comprises the following steps: deploying three or more robots on a work plane to form a group; the robots acquire mutual ranging values, fuse wheel odometer and inertial measurement unit displacement, and solve a positioning triangle with the ranging values and the displacement as constraints; a group adjustment optimization algorithm is used to fuse the triangular positioning results to obtain centimeter-level coordinates; the robots share trajectory coordinates and ranging values, and backup path memory; when a robot is abnormal, the remaining online robots acquire breakpoint positions through shared path memory and take over the work, and the remaining robots dynamically reorganize a triangular distribution network after excluding the abnormal robot; after the abnormal robot is online again, the online robots are used to measure the position of the abnormal robot and guide the abnormal robot to the breakpoint for resuming scanning. The application realizes multi-robot cooperative positioning without fixed facilities, path memory redundancy, and decentralized breakpoint takeover.

Description

Technical Field

[0001] This application relates to the field of robot localization and navigation technology, and in particular to a method and system for memorizing the localization path and performing breakpoint operations for a robot swarm. Background Technology

[0002] With the rapid development of the photovoltaic power generation industry, the installed capacity of large-scale photovoltaic power plants continues to grow, and the problem of dust and ash accumulation on the surface of photovoltaic modules seriously affects power generation efficiency. To replace traditional manual cleaning methods, various photovoltaic panel cleaning robots have emerged. In existing technologies, photovoltaic cleaning robots are typically equipped with a walking mechanism, cleaning brush components, a power supply module, and a basic control unit, enabling them to perform automated cleaning operations on the surface of photovoltaic arrays.

[0003] However, existing technologies still have the following technical shortcomings when applied to large-area photovoltaic panel cleaning scenarios: Existing positioning technologies struggle to balance the centimeter-level accuracy requirements of large-area photovoltaic (PV) arrays with deployment cost-effectiveness. On one hand, while SLAM solutions based on lidar or vision can achieve high-precision positioning and mapping, their sensors are expensive, computationally resource-intensive, and prone to feature matching failures in outdoor environments such as strong glare from PV panels and high-contrast shadows, leading to positioning drift. On the other hand, UWB or Bluetooth positioning solutions based on fixed base stations / tags require extensive pre-deployment of infrastructure around the PV array. For 100-megawatt PV power plants covering several square kilometers, the costs of base station deployment, power supply maintenance, and coordinate calibration are extremely high, and the base stations themselves are susceptible to single-point failures. While dead reckoning solutions are low-cost, wheeled odometers are prone to slippage and accumulated errors at the edges of PV panels and array gaps, resulting in positioning deviations reaching the meter level after prolonged operation, failing to meet the centimeter-level positioning requirements for fine cleaning.

[0004] Existing path memory mechanisms lack distributed redundancy, posing a risk of losing global information due to single-point failures. In the existing single-machine local storage mode, each robot only remembers its own trajectory. Once a robot stops due to power depletion, hardware failure, or communication interruption, its locally stored path memory, cleaned area boundaries, and uncleaned task assignment information are interrupted. Other robots cannot know the robot's work progress, which may result in cleaned areas being cleaned repeatedly and uncleaned areas being missed. Although cloud-based centralized storage can achieve data aggregation, in remote photovoltaic power station deployment scenarios, problems such as incomplete base station signal coverage, high network latency, and cloud service interruptions occur frequently. Once the communication link fails, the robot loses global map support and falls into a blind scanning state.

[0005] Existing breakpoint resume scanning technologies are limited to single-robot recovery or cloud-based scheduling and takeover, lacking decentralized breakpoint takeover capabilities within a robot ensemble. Current single-robot breakpoint resume scanning solutions require the offline robot to automatically recover and return to the breakpoint to continue operation. If the robot cannot recover for an extended period due to mechanical failure or battery depletion, the breakpoint area will remain uncleaned indefinitely. While existing cloud-based scheduling and takeover solutions can achieve multi-robot task transfer, they heavily rely on the real-time online status and global decision-making of cloud servers, rendering them ineffective in scenarios with limited communication. Existing technologies do not fully utilize the relative ranging and path memory sharing capabilities among online robots, failing to achieve a collaborative mechanism where a companion can directly take over the breakpoint and resume scanning.

[0006] Existing technologies do not fully utilize the spatial distribution characteristics of multiple robots in the same work area to suppress positioning errors. In scenarios where photovoltaic matrices are arranged in a regular pattern, multiple robots naturally form a measurable geometric distribution. However, existing technologies often treat each robot as an independent entity and only coordinate them at the task level through a server, without incorporating the real-time relative distance between robots as a positioning constraint into the pose calculation process. This results in the independent accumulation of positioning errors for each robot, making it impossible to cross-validate and suppress errors through group geometric relationships, and making it difficult to achieve centimeter-level positioning accuracy with low-cost sensor configurations.

[0007] For example, Chinese patent application CN120970617A discloses an intelligent navigation system for a sweeping robot. It provides environmental perception through a multi-source sensor array (LiDAR, depth vision sensor, collision sensor, and odometer), and achieves centimeter-level positioning by fusing SLAM calculation results and IMU data through a multimodal positioning module in a tightly coupled manner. However, it is a single-machine independent positioning architecture that relies on the high-cost SLAM sensor carried by the robot itself and does not involve mutual ranging and collaborative positioning between multiple robots.

[0008] For example, Chinese patent application CN116625373A discloses a positioning method and system for a self-propelled robot, which uses extended Kalman filtering fusion of GPS, IMU, and odometer for positioning and relies on pre-deployed RFID tags or boundary lines for position calibration. However, it is also a single-machine positioning architecture and is highly dependent on GPS satellite signals and pre-laid external facilities. Summary of the Invention

[0009] To address the aforementioned issues, this application provides a centimeter-level positioning path memory and breakpoint operation method and system adapted to self-propelled robot swarms. Through dynamic triangular distribution network collaborative positioning, distributed redundant path memory, and decentralized breakpoint takeover mechanism, it achieves multi-robot collaborative positioning and breakpoint operation without the need for fixed infrastructure.

[0010] The first aspect of this application provides a method for localization path memory and breakpoint operation of a robot swarm, including: Three or more self-propelled robots are deployed on the same working plane. Each self-propelled robot is equipped with a ranging module, a wheeled odometer and an inertial measurement unit. The three or more self-propelled robots constitute a robot group. Each of the self-propelled robots obtains the real-time mutual distance between itself and other robots in the group through the ranging module, and calculates the displacement increment of itself by fusing the wheel odometry and inertial measurement unit; using the real-time mutual distance and the displacement increment as constraints, the coordinates of each of the self-propelled robots are obtained by fusing the solution results of multiple positioning triangles through a group adjustment optimization algorithm. Each of the self-propelled robots shares and stores its own trajectory coordinates and the real-time mutual distance measurement values ​​in the robot group according to a time sequence, so that the path memory data of a single self-propelled robot can be backed up and called by other robots in the group; When any of the self-propelled robots in the group is determined to be in an abnormal state, the remaining online robots obtain the breakpoint location of the self-propelled robot determined to be in an abnormal state through the shared path memory data, and take over the work area corresponding to the breakpoint location to continue to perform the work task. At the same time, after the self-propelled robot determined to be in an abnormal state is removed, the remaining online robots dynamically reorganize the triangular distribution network. When the self-propelled robot that was determined to be in an abnormal state comes back online, the real-time mutual distance measurement value between the remaining online robots is used to calculate the current real position of the self-propelled robot that has come back online, and guides the self-propelled robot that has come back online to autonomously move to the breakpoint position to complete the continued scanning.

[0011] In an optional embodiment, each of the self-propelled robots collects its own coordinates at a preset frequency and shares them through group communication. Each of the self-propelled robots discretizes and stores trajectory points according to a preset precision grid. The storage structure includes a coordinate layer, a state layer, and a time layer. The coordinate layer is used to record the trajectory coordinates of the robot's center point. The state layer is used to mark the robot's motion state at each grid. The time layer is used to record the most recent update time of each grid state. The robot's motion states include a straight-line state, a turning state, and an edge-finding state. The state layer marks the grid corresponding to the turning state as a point trajectory to distinguish it from the linear trajectory corresponding to the straight-line state.

[0012] In an optional embodiment, each of the self-propelled robots establishes a local coordinate system with the fixed physical feature point of the working plane as the origin. Any three self-propelled robots are arranged into a positioning triangle. The side length constraint is established with the measured distance between each pair of the three self-propelled robots. At the same time, the kinematic constraint is established with the displacement increment of each self-propelled robot. The solution results of multiple positioning triangles are globally fused by the least squares adjustment method to eliminate the independent positioning cumulative error of each self-propelled robot and optimize the solution of the final coordinates of each self-propelled robot to meet the preset accuracy requirements.

[0013] In an optional embodiment, the ranging confidence level is determined based on the received strength of the ranging signal. The group adjustment optimization algorithm specifically involves: constructing a residual equation containing the side lengths of all positioning triangles; setting a weight matrix based on the ranging confidence level corresponding to each side length; and iteratively solving for the correction amount of each node coordinate using the weighted least squares method until the residual is less than a preset threshold.

[0014] In an optional embodiment, each of the self-propelled robots in the group broadcasts its own status after self-checking, or a designated online robot in the group acts as the master control node to comprehensively judge the status broadcast by each robot in order to identify abnormal status. The abnormal states include: when the turning angular velocity of the self-propelled robot exceeds a preset angular velocity threshold and the coordinate displacement per unit time is less than a preset displacement threshold, it is determined to be a spinning-in-place abnormality; when the difference between the displacement calculated by the wheel odometer and the displacement calculated by the inertial measurement unit continues to exceed a preset duration, it is determined to be a wheel slippage abnormality; when the self-propelled robot fails to send data packets for multiple consecutive communication cycles, it is determined to be a sudden offline abnormality; when the deviation between the coordinate change rate of the self-propelled robot and the average coordinate change rate of the entire group exceeds a preset multiple standard deviation and the positioning reliability exceeds a preset reliability threshold, it is determined to be a coordinate drift abnormality. The location reliability is a quantitative index that characterizes the reliability of the solution results of the population adjustment optimization algorithm.

[0015] In an optional embodiment, the re-connected self-propelled robot sends a recovery data packet containing its own identity and distance measurement values ​​of the online robots to other online robots via group communication. Each online robot constructs global constraints using real-time mutual distance measurement values, collects position data and distance measurement data from at least three online robots, establishes a set of constraint equations, solves for the current coordinates of the re-connected self-propelled robot using the least squares method, and then compares the current coordinates with the coordinates of the breakpoint position recorded when the self-propelled robot in an abnormal state went offline to obtain the position offset, guiding the re-connected self-propelled robot to the coordinates of the breakpoint position. When the re-connected self-propelled robot or any online robot in the group measures a distance error between itself and at least two online robots that is less than a preset accuracy threshold, the robot determines to complete the continued scanning.

[0016] In an optional embodiment, the dynamic reorganization of the triangular distribution network is performed as follows: When the number of remaining robots is greater than or equal to three after removing the self-propelled robots that are determined to be in an abnormal state, the positioning triangle is directly reconstructed using the remaining robots as nodes. When the number of remaining robots is two after removing the self-propelled robots that are determined to be in an abnormal state, the degenerate positioning mode is activated. Each self-propelled robot obtains the distance to the fixed boundary wall of the working plane through its own laser radar or ultrasonic sensor. This distance is used as a third constraint, and together with the real-time mutual distance measurement value between the two self-propelled robots, they form a positioning triangle to maintain two-dimensional positioning capability. When the number of remaining robots is one after removing the self-propelled robots that are determined to be in an abnormal state, the dead reckoning mode is switched and the operation speed is reduced to control the accumulated error. At the same time, a single-robot operation alarm is sent to the operation and maintenance terminal. One or more of the remaining online robots will take over the work area corresponding to the breakpoint and continue to perform the work tasks.

[0017] The second aspect of this application provides a robot swarm localization path memory and breakpoint operation system, including: A robot swarm consists of three or more self-propelled robots deployed on the same working plane. Each self-propelled robot is equipped with a ranging module, a wheeled odometer, an inertial measurement unit, and a communication module. The system constructs a dynamic triangular distribution network based on the real-time mutual ranging values ​​between the self-propelled robots as the basic architecture for group positioning and memory sharing. The positioning and calculation module is used to establish a local coordinate system with the fixed physical feature points of the working plane as the origin, and to obtain the coordinates of each self-propelled robot that meet the preset accuracy requirements by fusing the calculation results of multiple positioning triangles through a group adjustment optimization algorithm, based on the positioning triangles formed by any three self-propelled robots and the real-time distance measurement values ​​between each self-propelled robot and the dead reckoning displacement increment of each self-propelled robot as constraints. The path memory module is used to discretize and store the trajectory points of each self-propelled robot according to a preset precision grid, and to share the path memory data within the robot group so that the path data of a single self-propelled robot can be backed up and called by other robots in the group. An anomaly detection module is used to determine an anomaly status by broadcasting the status of each self-propelled robot after self-inspection or by comprehensive judgment by a designated master control node, record the breakpoint position and operation status of the self-propelled robot that is determined to be in an abnormal state, and remove the self-propelled robot that is determined to be in an abnormal state. The breakpoint recovery module is used to construct global constraints using real-time mutual ranging values ​​between other online robots after the self-propelled robot, which was determined to be in an abnormal state, is brought back online. This allows the module to calculate the current real position of the self-propelled robot that has been brought back online and guide it to the breakpoint position to continue scanning. The network reorganization module is used to dynamically reorganize the triangular distribution network of remaining robots after the self-propelled robot that has been determined to be in an abnormal state is removed.

[0018] In an optional embodiment, the ranging module is linked with the communication module to broadcast the ranging data at a communication cycle lower than the operation status reporting frequency, so as to reduce the channel load while ensuring the positioning reliability; the positioning calculation module is also used to periodically compare the angular velocity calculated by the wheel odometer and the inertial measurement unit with the angular velocity calculated based on the real-time mutual ranging value changes, and when the difference between the two exceeds a preset threshold, the ranging frequency of the ranging module is increased.

[0019] In an optional embodiment, the fixed physical feature points are the intersection points of the borders of the working plane and the bracket reference points.

[0020] Due to the adoption of the above technical solution, this application has at least one of the following beneficial effects compared with the prior art: 1. By constructing a dynamic triangular distribution network through mutual ranging between robots, and through the geometric constraints of multiple positioning triangles and the group adjustment optimization algorithm, centimeter-level positioning accuracy can be achieved without GPS signals, pre-deployed UWB base stations or RFID tags, or LiDAR or visual SLAM sensors. This solves the technical problems of high deployment cost, difficult maintenance, high risk of single-point failure, and easy failure of SLAM solutions in strong reflective environments in large-area photovoltaic matrix scenarios.

[0021] 2. The trajectory coordinates and mutual distance values ​​of each robot are shared and stored within the robot group. The path memory data of a single robot is backed up and accessed by other robots in the group. When any robot goes offline due to a malfunction, its path memory data is not lost, and other robots can obtain its work progress through shared data, thus solving the technical problem of global information loss caused by single-point failure.

[0022] 3. When a robot is identified as malfunctioning, the remaining online robots retrieve the breakpoint location using shared path memory data and automatically take over the cleaning task. This eliminates the need to wait for the malfunctioning robot to recover or rely on cloud server scheduling. Once the malfunctioning robot is back online, it can calculate its current position using distance measurements between other online robots, without needing to return to a fixed calibration point. This significantly improves the continuity and robustness of robot group operations in communication-constrained scenarios.

[0023] 4. By using motion feature vectors to determine abnormal states of the robot such as spinning in place, wheel slippage, sudden disconnection, and coordinate drift, and by adaptively reorganizing the triangular distribution network according to the number of remaining robots after the robot disconnects, the system ensures continuous operation capability under various abnormal conditions. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 A flowchart illustrating the robot swarm localization path memory and breakpoint operation method provided in an embodiment of this application; Figure 2 This is a schematic diagram of the framework of a robot swarm positioning path memory and breakpoint operation system provided in an embodiment of this application. Detailed Implementation

[0025] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It is understood that the specific embodiments described herein are only for explaining this application and not for limiting it. Furthermore, it should be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all structures. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0026] The terms "first," "second," etc., used in this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or apparatuses.

[0027] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0028] Existing photovoltaic cleaning robot positioning and navigation solutions typically employ a single-machine independent positioning architecture, relying on high-cost onboard sensors or pre-deployed external infrastructure. They do not fully consider the possibility of multiple robot groups achieving collaborative positioning through mutual ranging within the same working plane. Furthermore, traditional solutions lack a distributed redundancy mechanism for path memory and do not establish a decentralized takeover mechanism for resuming scanning after a breakpoint. This results in interruptions to work continuity or data loss when robots malfunction, communication is interrupted, or equipment is replaced, making it difficult to meet the requirements of high-precision positioning and robust breakpoint recovery for long-term outdoor operations of large-area photovoltaic matrices.

[0029] In view of this, this application provides a centimeter-level positioning path memory and breakpoint operation method and system adapted to self-propelled robot swarms. By introducing group cooperative localization based on dynamic triangular distribution networks, distributed redundant path memory, and decentralized breakpoint takeover mechanism, it achieves multi-robot cooperative localization and breakpoint operation without fixed infrastructure. Figure 1 As shown, Figure 1 A flowchart illustrating a robot swarm localization path memory and breakpoint operation method provided in an embodiment of this application includes: Three or more self-propelled robots are deployed on the same working plane. Each self-propelled robot is equipped with a ranging module, a wheeled odometer, and an inertial measurement unit. The three or more self-propelled robots form a robot group. In this embodiment, the working plane is the surface of the photovoltaic matrix. Three or more self-propelled cleaning robots are deployed on the same photovoltaic matrix. A local coordinate system is established with the fixed physical feature points of the photovoltaic matrix as the origin. The fixed physical feature points include the intersection of the right-angled frame of the photovoltaic array and the reference point of the array end support.

[0030] Each robot moves independently, acquiring real-time distance measurements between itself and other robots in the group via a ranging module. The robot's displacement increment is then calculated by fusing wheeled odometer measurements with an inertial measurement unit (IMU). The ranging module employs an ultra-wideband (UWB) ranging unit or a laser ranging unit, achieving a ranging accuracy of no more than 5 cm and covering the maximum diagonal distance of the entire photovoltaic array. The UWB module acquires relative distances through two-way ranging (TWR) or time difference of arrival (TDOA) methods, exhibiting strong resistance to multipath interference and suitability for the complex electromagnetic environment of the photovoltaic panel's metal frame. The wheeled odometer uses a high-resolution encoder, and the motion algorithm, combined with the motor and reduction wheel, controls the wheel rotation resolution to within 3°. Displacement is calculated through encoder pulse integration and acceleration / angular velocity integration. The IMU is an industrial-grade nine-axis IMU with absolute attitude angles.

[0031] Using real-time mutual distance measurements and displacement increments as constraints, the coordinates of each walking robot are obtained by fusing the solution results of multiple positioning triangles through a group adjustment optimization algorithm.

[0032] Each self-propelled robot establishes a local coordinate system with the fixed physical feature points of the working plane as the origin. Any three self-propelled robots are arranged into a positioning triangle. The side length constraints are established by the measured distance between each pair of the three self-propelled robots. At the same time, the kinematic constraints are established by the displacement increment of each self-propelled robot. The solution results of multiple positioning triangles are globally fused by the least squares adjustment method to eliminate the independent positioning cumulative error of each self-propelled robot and optimize the solution of the final coordinates of each self-propelled robot to meet the preset accuracy requirements.

[0033] Set up a robot and The real-time mutual distance between them is The displacement increment of each walking robot from the previous moment to the current moment is: Based on the trilateration method, establish the unknown coordinates of each walking robot. A system of constraint equations with variables:

[0034]

[0035]

[0036] in, For robots The unknown coordinates to be solved For robots The unknown coordinates to be solved For robots With robots Real-time mutual distance values ​​between them For robots The coordinates of the previous moment. For robots The displacement increment from the previous moment to the current moment is calculated by the fusion of wheel odometer and IMU; and All of these are robot IDs within the robot group, with values ​​ranging from 1 to 2. to , This represents the total number of robots in the robot group.

[0037] Based on the positioning triangle formed by any three robots, the least squares adjustment method is introduced to fuse the triangulation results; let there be a total Taiwan robots can form We have three positioning triangles, each providing a set of coordinate solutions, to construct an optimization objective function:

[0038]

[0039] in, To locate the triangle number, , From The number of combinations of any 3 self-propelled robots selected from a given number of robots. For the first The edge set of a positioning triangle. Indicates belonging to the first One side of a positioning triangle, For robots The dead reckoning coordinates are the robot's estimated current position calculated by integrating displacement increments from its previous position using only its onboard wheeled odometry and inertial measurement unit. They are not the final positioning result, but rather a kinematic constraint / prior value in the group adjustment optimization. The odometry constraint weight coefficient is used to balance the weights of ranging constraints and dead reckoning constraints in the optimization objective, solve the above nonlinear least squares problem, and obtain the coordinates of each walking robot; the above optimization problem is solved by the Gauss-Newton iteration method to improve the positioning accuracy.

[0040] Each self-propelled robot shares and stores its own trajectory coordinates and real-time mutual distance measurements within the robot group in a time-series format. This allows the path memory data of a single self-propelled robot to be backed up and retrieved by other robots in the group. Specifically, each self-propelled robot collects its own coordinates at a frequency of 5Hz to form a sequence of trajectory points.

[0041] in, For a sequence of trajectory points, For robot identification, ( , For robots The centimeter-level coordinates obtained at the current sampling time. For robots With robots The previous ranging value between the two samples at the current sampling time. For robots With robots The two distance measurements taken at the current sampling time. For robots The timestamp of the current sampling time For robots The motion state at the current sampling moment, For robots Location reliability at the current sampling time, subscript , … The sampling time number. The total number of trajectory points, index To work with robots Peer robots performing mutual ranging The number, The ellipsis indicates the continuation of the trajectory point sequence.

[0042] The trajectory points are discretized and stored in centimeter-level grid units. The storage structure includes a coordinate layer—recording the robot's center point trajectory in centimeter-level grids; a state layer—marking the robot's motion state for each grid, including straight, left turn, edge finding, and right turn (where left / right turn is marked as a point); and a time layer—recording the last update time of each grid state, used to determine the timeliness of the environment when resuming from a breakpoint.

[0043] When any autonomous robot in the group is determined to be in an abnormal state, the remaining online robots obtain the breakpoint location of the autonomous robot determined to be in an abnormal state through the shared path memory data, and take over the work area corresponding to the breakpoint location to continue to execute the work task. At the same time, after the autonomous robot determined to be in an abnormal state is removed, the remaining online robots dynamically reorganize the triangular distribution network.

[0044] The self-propelled robot constructs motion feature vectors locally by fusing data from wheeled odometry and IMU. ;in, Angular velocity, For acceleration, For linear velocity, This represents the rate of change of heading angle.

[0045] Each self-propelled robot in the group performs a self-check and broadcasts its own status, or a designated online robot in the group acts as the master control node to comprehensively judge the status broadcast by each self-propelled robot to identify abnormal states. When an abnormality is detected, the system records the current working position coordinates, records the current cleaning status, and removes the information of the self-propelled robot after an alarm is triggered to ensure accurate positioning of the remaining robot groups; the self-propelled robot determined to be in an abnormal state shuts down its roller brush motor, reduces power consumption, and awaits manual restoration.

[0046] When a self-propelled robot that was determined to be in an abnormal state comes back online, the real-time mutual ranging values ​​between the remaining online robots are used to calculate the current real position of the self-propelled robot that has come back online, and guide the self-propelled robot that has come back online to autonomously move to the breakpoint to complete the continued scanning.

[0047] The re-launched self-propelled robot sends a recovery data packet containing its own identification and distance measurement values ​​of the online robots to other online robots via group communication. Each online robot constructs global constraints using real-time mutual distance measurement values, collects position and distance measurement data from at least three online robots, establishes a set of constraint equations, solves for the current coordinates of the re-launched self-propelled robot using the least squares method, and then compares the current coordinates with the coordinates of the breakpoint recorded when the self-propelled robot in an abnormal state went offline to obtain the position offset, guiding the re-launched self-propelled robot to the coordinates of the breakpoint. When the distance error between the re-launched self-propelled robot and at least two online robots is less than a preset accuracy threshold, the robot determines to complete the continued scanning.

[0048] The re-launched self-propelled robot sends a recovery data packet containing its own identity and distance measurement values ​​of the online robots to other online robots through group communication. After the robot re-launches, it sends coordinates (0,0) as its initial coordinates. The position data and distance measurement data of at least three online robots are collected and a set of constraint equations are established.

[0049] By utilizing the distance difference between the walking robots, a global state matrix is ​​constructed locally. :

[0050] If a certain measured distance value The Euclidean distance calculated based on the self-reported coordinates of the two robots. If the deviation exceeds a preset threshold, a positioning anomaly is determined, and the system uses the measured distance value. The Euclidean distance calculated as a constraint correction between the self-reported coordinates of the two robots is used to adjust the Euclidean distance. :

[0051] in, Robots ,robot The centimeter-level coordinates obtained at the current sampling moment.

[0052] A system of linear equations is established for the three online robots. The current coordinates are solved by the least squares method. The current coordinates are then compared with the coordinates of the breakpoint recorded when the self-propelled robot, which was determined to be in an abnormal state, was taken offline to obtain the position offset. This position offset is the position offset of the self-propelled robot that was taken offline (whether it was moved or moved on its own) when it was put back online.

[0053] The current true position of the re-launched self-propelled robot is calculated through this displacement, rather than forcing it to return to the original breakpoint coordinates. The robot travels to the breakpoint position based on the wheel odometry and inertial measurement unit. Once the re-launched self-propelled robot or any online robot in the group measures a distance error between itself and at least two other online robots that is less than a preset accuracy threshold, the robot determines that it has completed the resume scanning function—that is, the resume scanning function is completed when the system measures a distance difference error between the coordinates and at least two other robots in the group that is less than 5cm.

[0054] After the eliminated robots are removed, the remaining robots dynamically reorganize the triangular distribution network: if the number of remaining robots is still greater than or equal to three, the positioning triangle is directly reconstructed with the remaining robots as nodes; if the number of remaining robots is two, the degenerate positioning mode is activated, using the fixed geometric features of the photovoltaic matrix (such as the array border line) as the third constraint to maintain the two-dimensional positioning capability; if only one robot remains, the dead reckoning mode is switched to, and the operation speed is reduced to control the accumulated error, while a single-machine operation alarm is sent to the operation and maintenance terminal.

[0055] In an optional embodiment, each walking robot collects its own coordinates at a preset frequency and shares them through group communication. Each walking robot discretizes and stores trajectory points according to a preset precision grid. The storage structure includes a coordinate layer, a state layer, and a time layer. The coordinate layer is used to record the trajectory coordinates of the robot's center point. The state layer is used to mark the robot's motion state at each grid. The time layer is used to record the most recent update time of each grid state. The robot's motion states include straight-line state, turning state, and edge-finding state. The state layer marks the grid corresponding to the turning state as a point trajectory to distinguish it from the linear trajectory corresponding to the straight-line state.

[0056] Each grid corresponds one-to-one with its physical location on the working plane. The grid size can be preset according to the required positioning accuracy. The trajectory points are discretized and stored in centimeter-level grid units. Each walking robot collects its own coordinates at a frequency of 5Hz to form a sequence of trajectory points. The coordinate layer records the robot's center point trajectory in centimeter-level grids. The state layer marks the robot's motion state for each grid, including straight, left turn, edge finding, and right turn (where left / right turn is marked as a point). The time layer records the last update time of each grid's state, which is used to determine the environmental timeliness when recovering from a breakpoint.

[0057] The ranging confidence level is determined based on the received strength of the ranging signal. The group adjustment optimization algorithm is as follows: construct a residual equation containing the side lengths of all positioning triangles; set a weight matrix according to the ranging confidence level corresponding to each side length; and iteratively solve the correction amount of each node coordinate using the weighted least squares method until the residual is less than the preset threshold.

[0058] Ranging confidence is determined based on the received signal strength (RSSI). Higher RSSI results in higher confidence for the ranging value and a larger weighting coefficient in the weighted least squares adjustment. The diagonal elements of the weighting matrix represent the normalized confidence values ​​for each ranging value, while the off-diagonal elements are zero. Alternatively, ranging confidence can be determined based on the signal-to-noise ratio (SNR) of the communication link or the variance of historical ranging values—higher SNR or smaller historical variance results in higher confidence.

[0059] In one optional embodiment, each walking robot in the group performs a self-check and broadcasts its own status, or a designated online robot in the group acts as the master control node to comprehensively judge the status broadcast by each walking robot in order to identify abnormal statuses.

[0060] Abnormal states include: when the turning angular velocity of an autonomous robot exceeds a preset angular velocity threshold and its coordinate displacement per unit time is less than a preset displacement threshold, it is judged as spinning in place; that is, when a certain autonomous robot's... And coordinate displacement within 2 consecutive seconds If the distance is less than 5mm, it is considered spinning in place.

[0061] When the difference between the displacement calculated by the wheel odometer and the displacement calculated by the inertial measurement unit continues to exceed a preset time, it is determined that the wheel is slipping abnormally; that is, when the displacement calculated by the wheel odometer... Displacement calculated by IMU If the difference between the values ​​is too large and continues for more than 3 sampling periods, it is determined that the wheel is slipping.

[0062] When an autonomous robot fails to send data packets for several consecutive communication cycles, it is determined to be a sudden offline anomaly; that is, when an autonomous robot fails to send data packets for N consecutive communication cycles (e.g., N=3), it is determined to be a sudden offline or a communication module failure.

[0063] When the deviation between the coordinate change rate of an autonomous robot and the average coordinate change rate of the entire group exceeds a preset multiple of the standard deviation, and the positioning reliability exceeds a preset reliability threshold, it is judged as an abnormal coordinate drift; that is, when the deviation between the coordinate change rate of a certain autonomous robot and the average coordinate change rate of the entire group exceeds 2 times the standard deviation, and the positioning reliability of the autonomous robot exceeds a preset reliability threshold, it is judged as an abnormal coordinate drift. When this occurs, it is determined to be coordinate drift.

[0064] Among them, location reliability is a quantitative index that characterizes the reliability of the solution results of the population adjustment optimization algorithm.

[0065] The dynamic reconfiguration of the triangular distribution network is as follows: When the number of remaining robots after removing the self-propelled robots judged to be in an abnormal state is greater than or equal to three, the positioning triangle is directly reconstructed using the remaining robots as nodes; when the number of remaining robots after removing the self-propelled robots judged to be in an abnormal state is two, a degenerate positioning mode is activated, and each self-propelled robot obtains the distance to the fixed boundary wall of the working plane through its own onboard lidar or ultrasonic sensor. This distance is used as a third constraint, and together with the real-time mutual distance measurement value between the two self-propelled robots, a positioning triangle is formed to maintain two-dimensional positioning capability; when the number of remaining robots after removing the self-propelled robots judged to be in an abnormal state is one, the dead reckoning mode is switched to and the working speed is reduced to control the accumulated error, while a single-machine operation alarm is sent to the operation and maintenance terminal; in this case, one or more of the remaining online robots take over the working area corresponding to the breakpoint location and continue to perform the work task.

[0066] If three or more robots remain, the positioning triangle is reconstructed using these remaining robots as nodes. If two robots remain, a degenerate positioning mode is activated, using the fixed geometric features of the photovoltaic matrix (such as the array border lines) as the third constraint to maintain 2D positioning capability. If only one robot remains, the dead reckoning mode is switched to, and the operation speed is reduced to control accumulated errors, while a single-robot operation alarm is sent to the maintenance terminal. Alternatively, in the two-robot mode, the third constraint can also be the distance to the fixed boundary wall measured by the robot's onboard LiDAR or ultrasonic sensors.

[0067] This application also provides a robot swarm localization path memory and breakpoint operation system, such as Figure 2 As shown, Figure 2 A schematic diagram of the framework of a robot swarm localization path memory and breakpoint operation system provided in an embodiment of this application includes: A robot swarm consists of three or more self-propelled robots deployed on the same working plane. Each robot is equipped with a ranging module, wheeled odometer, inertial measurement unit and communication module. The system constructs a dynamic triangular distribution network through the real-time mutual ranging values ​​between the robots to serve as the basic architecture for group positioning and memory sharing. The positioning and calculation module is used to establish a local coordinate system with the fixed physical feature points of the working plane as the origin. Based on the positioning triangle formed by any three self-propelled robots, and with the real-time distance measurement values ​​between each robot and the dead reckoning displacement increment of each robot as constraints, the module fuses the calculation results of multiple positioning triangles through a group adjustment optimization algorithm to obtain the coordinates of each robot that meet the preset accuracy requirements. The path memory module is used to discretize and store the trajectory points of each self-propelled robot according to a preset precision grid, and share the path memory data within the robot group so that the path data of a single self-propelled robot can be backed up and called by other robots in the group. The anomaly detection module is used to determine the anomaly status by broadcasting the status of each self-propelled robot after self-inspection or by comprehensive judgment by a designated master control node. It records the breakpoint position and operation status of the self-propelled robots that are determined to be in an abnormal state and removes the self-propelled robots that are determined to be in an abnormal state. The breakpoint recovery module is used to construct global constraints by using real-time mutual ranging values ​​between other online robots after an autonomous robot that has been judged to be in an abnormal state is brought back online. This allows the module to calculate the current real position of the re-brought-back autonomous robot and guide it to the breakpoint position to continue scanning. The network reorganization module is used to dynamically reorganize the triangular distribution network of the remaining robots after the self-propelled robots that have been determined to be in an abnormal state have been removed.

[0068] The ranging module works in conjunction with the communication module to broadcast ranging data at a communication cycle lower than the reporting frequency of the operation status, so as to reduce the channel load while ensuring the accuracy of the positioning. The positioning calculation module is also used to periodically compare the angular velocity calculated by the wheel odometer and the inertial measurement unit with the angular velocity calculated based on the real-time mutual ranging value changes. When the difference between the two exceeds a preset threshold, the ranging frequency of the ranging module is increased.

[0069] The fixed physical feature points are the intersection points of the working plane's borders and the support reference points.

[0070] In the several embodiments provided in this application, it should be understood that the disclosed methods and devices can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0071] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0072] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0073] The above description is merely an embodiment of this application and does not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A method for localization path memory and breakpoint operation in a robot swarm, characterized in that, include: Three or more self-propelled robots are deployed on the same working plane. Each self-propelled robot is equipped with a ranging module, a wheeled odometer and an inertial measurement unit. The three or more self-propelled robots constitute a robot group. Each of the self-propelled robots obtains the real-time mutual distance between itself and other robots in the group through the ranging module, and calculates the displacement increment of itself by fusing the wheel odometry and inertial measurement unit; using the real-time mutual distance and the displacement increment as constraints, the coordinates of each of the self-propelled robots are obtained by fusing the solution results of multiple positioning triangles through a group adjustment optimization algorithm. Each of the self-propelled robots shares and stores its own trajectory coordinates and the real-time mutual distance measurement values ​​in the robot group according to a time sequence, so that the path memory data of a single self-propelled robot can be backed up and called by other robots in the group. When any of the self-propelled robots in the group is determined to be in an abnormal state, the remaining online robots obtain the breakpoint location of the self-propelled robot determined to be in an abnormal state through the shared path memory data, and take over the work area corresponding to the breakpoint location to continue to perform the work task. At the same time, after the self-propelled robot determined to be in an abnormal state is removed, the remaining online robots dynamically reorganize the triangular distribution network. When the self-propelled robot that was determined to be in an abnormal state comes back online, the real-time mutual distance measurement value between the remaining online robots is used to calculate the current real position of the self-propelled robot that has come back online, and guides the self-propelled robot that has come back online to autonomously move to the breakpoint position to complete the continued scanning.

2. The method according to claim 1, characterized in that, Each of the aforementioned self-propelled robots collects its own coordinates at a preset frequency and shares them through group communication. Each of the aforementioned self-propelled robots discretizes and stores trajectory points according to a preset precision grid. The storage structure includes a coordinate layer, a state layer, and a time layer. The coordinate layer is used to record the trajectory coordinates of the robot's center point. The state layer is used to mark the robot's motion state at each grid. The time layer is used to record the most recent update time of each grid state. The robot's motion states include straight-line state, turning state, and edge-finding state. The state layer marks the grid corresponding to the turning state as a point trajectory to distinguish it from the linear trajectory corresponding to the straight-line state.

3. The method according to claim 1, characterized in that, Each of the self-propelled robots establishes a local coordinate system with the fixed physical feature point of the working plane as the origin. Any three self-propelled robots are arranged into a positioning triangle. The side length constraint is established by the measured distance between each pair of the three self-propelled robots. At the same time, the kinematic constraint is established by the displacement increment of each self-propelled robot. The solution results of multiple positioning triangles are globally fused by the least squares adjustment method to eliminate the independent positioning cumulative error of each self-propelled robot and optimize the solution of the final coordinates of each self-propelled robot to meet the preset accuracy requirements.

4. The method according to claim 1, characterized in that, The ranging confidence level is determined based on the received strength of the ranging signal. The group adjustment optimization algorithm is as follows: construct a residual equation containing the side lengths of all positioning triangles; set a weight matrix according to the ranging confidence level corresponding to each side length; and iteratively solve the correction amount of each node coordinate using the weighted least squares method until the residual is less than a preset threshold.

5. The method according to claim 1, characterized in that, Each self-propelled robot in the group broadcasts its own status after self-checking, or a designated online robot in the group acts as the master control node to comprehensively judge the status broadcast by each robot in order to identify abnormal status. The abnormal states include: when the turning angular velocity of the self-propelled robot exceeds a preset angular velocity threshold and the coordinate displacement per unit time is less than a preset displacement threshold, it is determined to be a spinning-in-place abnormality; when the difference between the displacement calculated by the wheel odometer and the displacement calculated by the inertial measurement unit continues to exceed a preset duration, it is determined to be a wheel slippage abnormality; when the self-propelled robot fails to send data packets for multiple consecutive communication cycles, it is determined to be a sudden offline abnormality; when the deviation between the coordinate change rate of the self-propelled robot and the average coordinate change rate of the entire group exceeds a preset multiple standard deviation and the positioning reliability exceeds a preset reliability threshold, it is determined to be a coordinate drift abnormality. The location reliability is a quantitative index that characterizes the reliability of the solution results of the population adjustment optimization algorithm.

6. The method according to claim 1, characterized in that, The re-launched self-propelled robot sends a recovery data packet containing its own identity and distance measurement values ​​to other online robots via group communication. Each online robot constructs global constraints using real-time mutual distance measurement values, collects position and distance measurement data from at least three online robots, establishes a set of constraint equations, solves for the current coordinates of the re-launched self-propelled robot using the least squares method, and then compares the current coordinates with the coordinates of the breakpoint recorded when the self-propelled robot in an abnormal state went offline to obtain the position offset, guiding the re-launched self-propelled robot to the coordinates of the breakpoint. When the re-launched self-propelled robot or any online robot in the group measures a distance error between itself and at least two online robots that is less than a preset accuracy threshold, the robot determines to complete the continued scanning.

7. The method according to claim 1, characterized in that, The method for dynamically reorganizing the triangular distribution network is as follows: When the number of remaining robots is greater than or equal to three after removing the self-propelled robots that are determined to be in an abnormal state, the positioning triangle is directly reconstructed using the remaining robots as nodes. When the number of remaining robots is two after removing the self-propelled robots that are determined to be in an abnormal state, the degenerate positioning mode is activated. Each self-propelled robot obtains the distance to the fixed boundary wall of the working plane through its own laser radar or ultrasonic sensor. This distance is used as a third constraint, and together with the real-time mutual distance measurement value between the two self-propelled robots, they form a positioning triangle to maintain two-dimensional positioning capability. When the number of remaining robots is one after removing the self-propelled robots that are determined to be in an abnormal state, the dead reckoning mode is switched and the operation speed is reduced to control the accumulated error. At the same time, a single-robot operation alarm is sent to the operation and maintenance terminal. One or more of the remaining online robots will take over the work area corresponding to the breakpoint and continue to perform the work tasks.

8. A robot swarm localization path memory and breakpoint operation system, characterized in that, include: A robot swarm consists of three or more self-propelled robots deployed on the same working plane. Each self-propelled robot is equipped with a ranging module, a wheeled odometer, an inertial measurement unit, and a communication module. The system constructs a dynamic triangular distribution network based on the real-time mutual ranging values ​​between the self-propelled robots as the basic architecture for group positioning and memory sharing. The positioning and calculation module is used to establish a local coordinate system with the fixed physical feature points of the working plane as the origin, and to obtain the coordinates of each self-propelled robot that meet the preset accuracy requirements by fusing the calculation results of multiple positioning triangles through a group adjustment optimization algorithm, based on the positioning triangles formed by any three self-propelled robots and the real-time distance measurement values ​​between each self-propelled robot and the dead reckoning displacement increment of each self-propelled robot as constraints. The path memory module is used to discretize and store the trajectory points of each self-propelled robot according to a preset precision grid, and to share the path memory data within the robot group so that the path data of a single self-propelled robot can be backed up and called by other robots in the group. An anomaly detection module is used to determine an anomaly status by broadcasting the status of each self-propelled robot after self-inspection or by comprehensive judgment by a designated master control node, record the breakpoint position and operation status of the self-propelled robot that is determined to be in an abnormal state, and remove the self-propelled robot that is determined to be in an abnormal state. The breakpoint recovery module is used to construct global constraints using real-time mutual ranging values ​​between other online robots after the self-propelled robot, which was determined to be in an abnormal state, is brought back online. This allows the module to calculate the current real position of the self-propelled robot that has been brought back online and guide it to the breakpoint position to continue scanning. The network reorganization module is used to dynamically reorganize the triangular distribution network of remaining robots after the self-propelled robot that has been determined to be in an abnormal state is removed.

9. The system according to claim 8, characterized in that, The ranging module is linked with the communication module to broadcast ranging data at a communication cycle lower than the reporting frequency of the operation status, so as to reduce the channel load while ensuring the accuracy of the positioning. The positioning calculation module is also used to periodically compare the angular velocity calculated by the wheel odometer and the inertial measurement unit with the angular velocity calculated based on the real-time mutual ranging value changes. When the difference between the two exceeds a preset threshold, the ranging frequency of the ranging module is increased.

10. The system according to claim 8, characterized in that, The fixed physical feature points are the intersection points of the borders of the working plane and the reference points of the support.

Images